Postman × Gupshup · Demo Story

Adopt & scale AI.
Stay in control.

Use cases where Postman helps Gupshup adopt AI fast, while keeping governance, cost, and reliability in check.

LLM adoption Security Cost
~45 MIN 2 ACTS GATEWAY → AI ENGINEER
Scroll
00

The goals

Act 1

Adopt & scale AI

Go from AI pilots to platform-wide adoption.

Act 1

Governance

Central control and uniform policy over every model and tool.

Act 1

Cost control

Predictable AI spend you can plan and defend.

Act 2

Ship with confidence

Evolve a complex landscape safely.

Both

Full visibility

One live view of every service and dependency.

Act 01 · The Use Case
Safe acceleration & scaling of AI adoption

Let enterprise dev teams rapidly adopt and scale AI agents, MCP servers, and LLMs, without a "wild west" of rogue tooling. Developers get instant, self-serve access. Security, Compliance, and FinOps get central control, one set of policies, and full spend visibility, all through a single control plane.

Our proposed solution→Fabric Gateway
The point
Adopt and scale AI, and stay in control.
The idea
Every AI call goes through one gate you own.
The story
A dev points to one URL and gets everything. You set the rules. Secrets stay home.
The wow
Flip a switch → the same call is blocked. Live, on screen.
Fixes
Hidden AI use, leaked keys, surprise bills.
How it worksRules live outside · your data stays in your cloud
POSTMAN CLOUD · THE RULES Gateway settings who · limits · models rules sync down ↓ YOUR CLOUD (VPC) Settings copy Your app Claude Code · agents just a gate pass no real keys GATEWAY Check access who can use what key vault RULES Usage limit tokens · rate Block 429 pick model · allow / block · track cost OUTSIDE Anthropic · Claude the AI model Tools New Relic · Tavily · … Keys never leave your cloud the app only has a gate pass
01

Easy adoption

  • One URL for every tool
  • Find tools, don't hardcode them
  • One key for Slack, Jira, GitHub
02

Governance

  • Decide who gets what
  • Allow or deny any tool
  • Keys never leave your cloud
03

Cost control

  • Budgets and token limits
  • Route to the cheapest model
  • See spend live
DemoWhat's on screen
01

Set up

Quick look at the settings. 15 sec.

→
02

Ask the agent

One URL → answers from Claude + a tool.

→
03

Change a rule

Lower the limit in the gateway.

→
04

Blocked

Same request, now stopped. Live.

→
05

Find tools

100s of tools behind one door.

Devs move day one
No keys on laptops
Spend capped
Everything logged
Back pocketPassport does the same thing. Save it for if they ask.
ZOOM OUT

Now the leader sees

The same traffic gives you the big picture. Nothing extra to set up.

248

Count

Every service across all teams. Counted and owned.

Scores

See which teams do well, over time.

Risk

✓✓!✓✓!✓✓

Find hidden APIs no one is tracking.

Act 02 · The Use Case
Ship changes fast without breaking production

Give engineering leaders a live, architect-level view of every service and how they connect. Teams can see the blast radius of any change, find exactly where to make it, and catch breaking changes before they ship, across internal services and outside partners like Meta and WhatsApp.

Our proposed solution→AI Engineer
The point
It sees the whole map of your services.
The idea
Ask it what breaks, where to change, what's safe to ship.
The story
"Meta changes a WhatsApp field." Ask once → see what breaks and plan the fix.
The wow
The live map on screen. Then a review that catches breaks people miss.
Fixes
"Where do I make this change?" · surprise outages · shallow reviews.

The questions a leader wants answered

Q1

What outside services do we depend on, and who owns them?

Q2

If Meta changes something, what breaks and how big is the fix?

Q3

Will today's change break another team?

Q4

Do our API docs still match the code?

Q5

What could quietly take down production?

Q6

Which services are dead weight?

The map · answers Q1 & Q2

One change → what it hitsMeta / WhatsApp → your services
Meta / WhatsApp outside · field changed msg-router templates webhooks billing notify analytics cx-inbox ledger reports breaks knock-on ask once, see it all
  AI Engineer · Ask
you ▸

What outside services do we rely on, and where's the risk?

◆ AI Engineer
We rely on 11 outside services, across messaging, AI, voice, and payments. The biggest risk is Meta / WhatsApp. Three services use it directly, and about nine more sit behind those. A change there hits the most.
Meta · WhatsAppmessaging · 3 direct
Google · RCSmessaging
Apple · Messagesmessaging
Amazon Bedrockai
Anthropic · Claudeai
OpenAIai
Deepgramspeech
ElevenLabsvoice
Razorpaypayments
Twiliosms
SendGridemail

The auto review · answers Q3, Q4 & Q5

Every pull request gets checked. Not just "is the code clean," but "who does this break, and is prod safe?"

PR #545 · API Impact Review

Add a source field to tasks

⚠ Nothing breaks · 2 things to fix first
Safe
Users are fine.The public routes don't expose the new field.
Changed
The API changed on /tasks.A new source field now goes over the wire.
Docs
The docs no longer match the code.Spec + Postman collection weren't updated. Fixes drafted, not applied.
Callers
Checked every caller on the map. Two services, and neither breaks.One ignores the field, the other doesn't use these routes.
Danger
A database change has to ship first, or prod fails.Wrong order and you get Unknown column 'source'. Check before merge.
Nothing changed · all of this is a suggestion

In short, it does three things

"If Meta drops this, what breaks?"

Shows the fallout

See every service a change hits. Plan before it lands.

"Where do I make this change?"

Points the way

Give it a bug, it points to the right service. No more hunting.

"Docs drifted, and prod could fail."

Reviews every PR

Catches breaks across services that people miss.

+ Builds a mock service so teams don't wait on each other.
DemoWhat's on screen
01

The map

Whole landscape on screen. It lands instantly.

→
02

Ask it

"If Meta changes this…" → services light up.

→
03

Review posts

PR #545 gets checked, risks flagged.

Plan changes, don't fear them
Fix bugs faster
Fewer surprise outages
03 · THE PAYOFF

Back to your goals

Two products, one picture, mapped to what you wanted.

Your goals → answered

Adopt & scale AIGateway
GovernanceGateway
Cost controlGateway
Ship with confidenceAI Engineer
Full visibilityCatalog

One system, three parts

ControlGateway. One door, all the locks.
InsightAI Engineer. Sees the whole map.
VisibilityCatalog. Count, scores, risk.
All on Postman.

Go fast. Stay safe.

Postman × Gupshup · Demo Story